Goldman Sachs: AI Default Risk Has Tripled Since Early 2026
nashnova research
Goldman Sachs reports that AI-related credit default risk has tripled since January, with CDS spreads on Oracle, CoreWeave, Nvidia and Broadcom at or near all-time wides — the credit market is pricing in a message: AI's bill is coming due.
Default risk tripled — what exactly happened?
Goldman's credit strategy team flags AI-related credit default risk as having tripled since January 2026.
CDS — credit default swaps, essentially insurance contracts against bond defaults — on Oracle, CoreWeave, Nvidia and Broadcom are all at or near all-time wide levels. This means → the market has never priced the odds of these companies defaulting this high.
Strategist Reid Zhou quoted a London client to capture the mood: "It feels like a crisis without a crisis." In plain terms = asset prices are under broad pressure, but no actual defaults have hit — and nobody knows which way to lean.
AI has rattled credit markets three times this year — each time differently?
AI triggered three distinct credit scares in 2026, each with a different transmission mechanism: February hit private credit exposure to AI disruption; July hit new debt supply from hyperscale cloud firms; the current wave focuses on whether companies can bear higher funding costs.
This reflects a deepening pattern of concern — from "who gets disrupted" to "who borrowed too much" to "who can repay."
Goldman sees more to come. The next worry centers on demand and ROI — whether capex converts to returns fast enough. Amazon burning $33.5 billion in free cash flow in a single quarter while adding $48.1 billion in new debt is cited as the defining case study.
How bad is it for the weakest borrowers?
Broad indices have pulled back from their widest 2026 levels: CDX IG and CDX HY fell from the 69th and 74th percentiles to the low 50s. But USD CCC-rated — the lowest-quality high-yield — cash spreads sit at the 96th percentile, far above the 51st percentile at their earlier 2026 wides.
In plain terms = the broad market has healed significantly, but the weakest issuers never joined the recovery — they kept sinking.
The numbers: CCC spreads hit 1,157 basis points, widening 338 bps year-to-date and 195 bps in the past three months alone; the overall high-yield index widened just 20 bps over the same period. This means → risk is concentrating in the most fragile corner of the market.
Which sectors got hit hardest — and which became safe havens?
In investment grade, AI-linked sectors bear the most pressure: tech widened 50 bps YTD to 144 bps, telecom and media each widened 44 bps. Healthcare widened just 1 bp, industrials 4 bps, and oil & gas actually tightened 10 bps.
High-yield divergence is even sharper: HY telecom spreads widened 219 bps YTD to 599 bps; HY oil & gas tightened 108 bps, with crude near $100 a barrel turning energy issuers into the market's de facto safe asset.
This reflects a clear pattern: the deeper an industry's ties to AI capex, the greater the credit stress; industries anchored to physical commodities have become the capital refuge.
What are investors doing — and why is this the "maximum pain" setup?
Goldman describes London credit investors as largely directionless, seeing too many factors that could push prices either way. Positioning is light across the board.
The consensus trade: long equities, short credit. Credit — cash bonds especially — is seen as an asymmetric short at current valuations.
Yet nearly everyone admits a year-end rally is the most likely outcome — and almost nobody has positioned for it. In plain terms = everyone expects a bounce, but no one is willing to buy — that is precisely how "maximum pain" forms: whichever way the market moves, most participants end up on the wrong side.
Has the market's pricing logic shifted — from "broad brush" to "name by name"?
Strategist Reid Zhou emphasizes that the market is moving from pricing AI credit risk as a bloc to differentiating issuer by issuer: "Every earnings report and every new bond deal is separating issuers — we expect this trend to continue."
This means → beneath the surface stability of broad indices, internal divergence keeps widening. The index looks calm; individual credits tell a different story.
This signals that credit markets have entered a new phase: the macro narrative is fading, and individual credit quality is becoming the sole pricing anchor.
市场有风险，内容仅供研究参考，不构成投资建议。